Tidal power has the potential to play a significant role in future renewable energy generation. To achieve that potential, developers want to use larger and highly reliable blades on tidal turbines, to produce greater quantities of cost-efficient energy.
One big challenge is the difficulty in replicating under-sea conditions when testing tidal technology – to understand how blades will cope in a complex and variable environment.
This is being addressed by data scientist Sergio Lopez in his work with FASTBLADE, an innovative new test facility at the Babcock International site at Rosyth, Fife.
FASTBLADE will recreate the force of the ocean on tidal turbine blades in the lab, with the potential to exert 20 years of underwater forces on a turbine within just three months – to discover if a blade can withstand the load and if not, where improvements are needed.
Data is at the very heart of the project and Lopez’s research could open new pathways for data scientists to get involved in tidal blade testing – and in testing a wider range of composite or metal structures in different industries, including aerospace and both land and sea transport.
His work as a Marie Curie Fellow is supported by the European Union Horizon 2020 fund and the Data-Driven Innovation initiative in Edinburgh, working with several industry collaborators.
Lopez trained as a civil engineer and quickly began to focus on big data simulations, in both his native Guatemala and Italy, before he came to Scotland.
He says the combination of big data analysis, with an innovative approach to re-using energy, makes FASTBLADE stand out. ‘Regenerative pumping’ means energy is stored between test cycles, like an electric car, then re-used. This means FASTBLADE can test bigger blades and apply more complex loads quickly and effectively, making testing cheaper as it is more efficient.
“We can test faster than other centres and will produce a lot of data from many different sensors and systems.” says Lopez, who has been working on the three-year project since November 2020 and is prepared for when testing can start in late 2021.
One historic challenge has been a limited amount of reliable data from the ocean to help design blades that can last for up to 20 years in a harsh environment.
“For now, we are focusing on Scottish tidal data, collected by the University of Edinburgh in the Redapt Project at the European Marine Energy Centre (EMEC) in Orkney,” says Lopez. “We analyse ocean data recorded every second for many months. We extend that to 20 years, and finally to loads that FASTBLADE can replicate in three months.”
Lopez admits it is extremely complex to replicate environmental loads and conditions in a test facility.
“The challenges are multiple,” he says. “The first thing is the data available to know how to define the loads, then to attempt to replicate all of them. This process, with traditional methods, is highly computationally expensive and takes a long time. We’re developing simpler, faster, data-based models, to reduce the computational time, allow us to perform more simulations and get more accurate results.”
Rather than one-speed, up-and-down movements, data-driven inputs allow FASTBLADE to move the turbine blade at different speeds and positions, so it experiences more realistic and variable loads.
Data from these tests, using more than 200 different sensors, will allow the blade to be analysed under more realistic conditions in a controlled environment.
“We need to control everything in real-time, detect changes in the blade structure,” says Lopez. “Under the sea, you cannot control the flow or stop the test if something goes wrong. .”
Previous research relied largely on trial and error inputs. Now data analysis, married with machine learning, is improving the process hugely. Lopez explains: “Data-driven techniques allow us to quickly and easily make adjustments to many different settings, to get the outcomes we want.
“Machine learning can improve those trial-and-error approaches, finding out which set-ups work, identifying complex interactions that we do not have exact solutions for, or where computational costs are exceptionally high. A standard computer could take months to solve the problem by traditional methods, but it could take minutes with machine learning. So we are designing and training these new algorithms.”
Jeff Steynor, a Project Manager and Senior Experimental Officer at the University of Edinburgh, who is supervising Lopez’s work, says: “This approach means you will be able to detect failure quickly; if you have a sensor on a tidal blade in the ocean, and you can get an alert saying the blade is about to fail, you can mobilise a team more quickly to remove the blade in a controlled environment and replace it. The worst thing would be to lose a blade [without warning] because the other blades on that tidal turbine are also likely to fail, and you’re replacing three blades, not one. The ability to predict when that’s going to happen is extremely important so you can intervene in a cost-effective and planned manner.”
Lopez says one large project challenge is the amount of data being produced – already about 500 Gigabytes every day. “I’m looking to streamline that data – reducing test times by getting more from all the equipment in real-time to make FASTBLADE more efficient,” he says.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 801215.
Read the latest Case studies
From assessing children’s access to health centres in Uganda to understanding adolescents’ use of social…
It has been estimated the UK will need to increase its number of electric vehicle…